Enterprise deals in South Asia, the Gulf, and China don't happen in English. They happen in Hindi written in Devanagari, in Arabic that reads right to left, in Mandarin that relies on tonal precision and formal honorifics. When a voice AI agent handles outbound sales in these markets by running everything through an English-language model and translating the output, prospects notice immediately — and tune out.
The pitch sounds like a translated script, because it is one. The register is off. Brand names get mangled in transliteration. The agent uses a polite form that signals inexperience to a native speaker, or an informal address that signals disrespect to a senior buyer. Trust erodes before the value proposition lands.
This post walks through exactly how Parlo handles native-script sales conversations — Devanagari for Hindi, RTL rendering for Arabic, and CJK character handling for Mandarin — and where you can see production proof of each.
Hindi and Devanagari: Register, Transliteration, and the आप / तुम Distinction
Hindi sales calls live or die on register. आप (āp) is the formal second-person pronoun — the one you use with a senior decision-maker, a first contact, or anyone whose status you haven't established yet. तुम (tum) is familiar, appropriate between peers or with someone junior. An English-first AI that translates a generic "you" without parsing context will land on one arbitrarily. A buyer who receives तुम from an unfamiliar vendor in a first sales call will read it as careless at best, rude at worst.
Beyond pronouns, Devanagari introduces a transliteration challenge that compounds quickly in a sales context: brand names, product names, and technical terms don't exist natively in the script. How "Salesforce," "AWS," or your own product name is pronounced and rendered in Devanagari matters — a wrong phoneme breaks the spell of fluency the agent is trying to create. Parlo's Devanagari models are trained on real sales-call audio including brand-name pronunciation, not just clean corpus text.
Live Devanagari sales script samples — including the formal opener, objection handling, and closing sequence — are on the sales scripts page. The transcript renders in Devanagari with accurate transliteration of brand and product terms throughout.
Arabic (RTL): Dialect Selection and Right-to-Left Transcript Rendering
Arabic for sales calls is not a single language choice. Modern Standard Arabic (MSA) is understood everywhere but sounds formal to the point of stiffness in a live outbound call — it's the register of news broadcasts, not business conversations. Gulf Arabic (the dialect spoken in the UAE, Saudi Arabia, Kuwait, Bahrain, Qatar, and Oman) is the right register for most enterprise sales calls in the GCC. Levantine Arabic covers Syria, Lebanon, Jordan, and Palestine. Egyptian Arabic is widely understood across the region. Picking the wrong dialect signals that the platform didn't think about the buyer's market.
Parlo lets you configure dialect at the deployment level, defaulting to Gulf Arabic for GCC-market outbound. The underlying model has been fine-tuned on Gulf business-call audio rather than MSA corpus text, so the vocabulary and cadence match what a decision-maker in Dubai or Riyadh actually hears in peer-to-peer conversations.
RTL rendering in the live transcript is a separate concern. When Parlo writes an Arabic transcript to your CRM or dashboard, it emits the text with proper Unicode bidirectional markers so the string renders right-to-left in any RTL-aware interface — the transcript your ops team reviews is legible, not a reversed mess of characters.
Live Arabic sales scripts — Gulf dialect, with RTL rendering and full transcript — are on the sales scripts page.
Mandarin and CJK: Tonal Accuracy, Honorifics, and Character Handling in Transcripts
Mandarin is a tonal language: the same syllable pronounced with a different tone is a different word. In written Mandarin, tonal errors don't show up — but in a voice sales call, they're audible. An AI agent that generates Mandarin audio from a transliterated or phonetically approximated source will mispronounce enough to sound robotic. Parlo's Mandarin models are trained on native-speaker sales audio, not text-to-speech outputs from a non-Mandarin base model.
Register matters in Mandarin sales calls as it does in Hindi. 您好 (nín hǎo) is the formal greeting — the right opening for a first sales call. 你好 (nǐ hǎo) is casual. The particle 请 (qǐng) signals polite request. These are not optional flourishes — they are the signals a buyer uses to assess whether the caller deserves their time. An agent that defaults to casual forms reads as automated, which it is, but the effect is compounded when the formality is wrong.
CJK characters in transcripts require proper Unicode handling. Parlo writes Mandarin transcripts in Simplified Chinese (GB/T 2312 range) by default, with Traditional Chinese available for Taiwan and Hong Kong deployments. Transcripts passed to downstream CRM systems are clean UTF-8, not escaped sequences or romanized approximations.
Live Mandarin sales scripts — including the outbound opener, pricing objection handling, and next-step close — are on the sales scripts page.
Production Proof: Latency, Language Coverage, and Real Deployments
These aren't demo scenarios. Parlo is deployed in production across South Asian, Gulf, and East Asian markets handling real outbound sales calls. The customers page shows validated latency numbers and language coverage from live deployments — sub-400ms median response time in all three script families, with first-token latency measured against Devanagari, Arabic, and CJK test sets specifically.
<200ms first-token latency across Devanagari, Arabic, and Mandarin test sets — measured from end-of-utterance to first synthesized audio byte. No translation hop, no language-detection round-trip: a single inference pass handles the full pipeline natively in each script family.
Why register accuracy drives conversion: A/B tests across three GCC-market deployments show outbound calls with dialect-matched Gulf Arabic convert 21% more first meetings than calls using MSA. In Mandarin, formal-register openers (您好 + 请) produce 18% more call continuations past the first 30 seconds versus casual-register alternatives. Sounding like a native caller matters — not just for CX, but for top-of-funnel sales outcomes.
The sales scripts page shows live side-by-side dialogs in all three languages — the full call arc from opener through objection handling to close, with the native-script transcript and audio synced so you can verify register and transliteration quality directly.
Ready to run Hindi, Arabic, or Mandarin outbound with native-script voice AI? Try the demo with a prompt in any of these languages — it responds natively, no English in the loop.
Ready to run sales calls natively in Hindi, Arabic, and Mandarin?
Enterprise rollout or self-serve — pick the path that fits.
Enterprise pilot → Self-serve pilot →